Re: DMD now incorporates a disassembler
On Sunday, 9 January 2022 at 06:04:25 UTC, max haughton wrote: On Sunday, 9 January 2022 at 02:58:43 UTC, Walter Bright wrote: I've never seen one. What's the switch for gcc to do the same thing? For GCC/Clang you'd want -S (and then -masm=intel to make the output ~~beautiful to nobody but the blind~~ readable). I prefer -save-temps -fverbose-asm which generates a supplemental .i and .s file without changing the .o file.
Re: DMD now incorporates a disassembler
On 1/9/2022 11:33 AM, max haughton wrote: https://stackoverflow.com/questions/2511018/how-does-objdump-manage-to-display-source-code-with-the-s-option obj2asm does the same thing: https://www.digitalmars.com/ctg/obj2asm.html
Re: DMD now incorporates a disassembler
On 1/8/2022 10:04 PM, max haughton wrote:
For GCC/Clang you'd want -S
I know about that, but take a look at it:
> cat fred.c
int fred(int a[10])
{
return a[11];
}
> cc -S test.c
> cat test.s
.file "test.c"
.text
.globl test
.type test, @function
test:
.LFB0:
.cfi_startproc
pushq %rbp
.cfi_def_cfa_offset 16
.cfi_offset 6, -16
movq%rsp, %rbp
.cfi_def_cfa_register 6
movl$0, %eax
popq%rbp
.cfi_def_cfa 7, 8
ret
.cfi_endproc
.LFE0:
.size test, .-test
.ident "GCC: (Ubuntu 4.8.4-2ubuntu1~14.04.4) 4.8.4"
.section.note.GNU-stack,"",@progbits
Contrast with what -vasm does:
> cat test.d:
int fred(int* a)
{
return a[11];
}
> dmd -c test.d -vasm
_D4test4fredFPiZi:
: 8B 47 2Cmov EAX,02Ch[RDI]
0003: C3 ret
***
-vasm gives me what I want to see. There aren't extra steps to getting it, the
object code is included, and all the boilerplate is omitted.
It's all about the friction.
Re: DMD now incorporates a disassembler
On 1/8/2022 10:04 PM, max haughton wrote: Anyway, I've been playing with -vasm and I think it seems pretty good so far. There are some formatting issues which shouldn't be hard to fix at all (this is why we asked for some basic tests of the shape of the output), put I think I've only found one (touch wood) situation where it actually gets the instruction *wrong* so far. Testing it has led to me finding some fairly bugs in the dmd inline assembler, which I am in the process of filing. Thanks. This helps a lot!
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 23:14:54 UTC, Dukc wrote:
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote:
Compile with -vasm to see it! Enjoy!
For the file test.d:
int demo(int x)
{
return x * x;
}
Compiling with:
dmd test.d -c -vasm
prints:
_D4test4demoFiZi:
: 89 F8 mov EAX,EDI
0002: 0F AF C0imulEAX,EAX
0005: C3 ret
https://github.com/dlang/dmd/pull/13447
Wow, very useful! This feature surely lowers the bar to check
the disassembly when optimising. Thanks!
I'm slightly disappointed it does not output the asm inlined to
D code but that's just my daydreaming with no practical reasons
to back it up.
https://stackoverflow.com/questions/2511018/how-does-objdump-manage-to-display-source-code-with-the-s-option
Enjoy
Re: DMD now incorporates a disassembler
On Saturday, 8 January 2022 at 20:50:56 UTC, max haughton wrote: Most other compilers have been able to do this for years. Forever. I have never used a C compiler that doesn't output assembly on request. Pretty much a cultural requirement as C compilers used to pipe asm through a separate assembler.
Re: DMD now incorporates a disassembler
On Sunday, 9 January 2022 at 02:58:43 UTC, Walter Bright wrote:
I've never seen one. What's the switch for gcc to do the same
thing?
For GCC/Clang you'd want -S (and then -masm=intel to make the
output ~~beautiful to nobody but the blind~~ readable). This
dumps the output to a file, which isn't exactly the same as what
-vasm does, but I have already begun piping the -vasm output to a
file since (say) hello world yields a thousand lines of output
which is much easier to consume in a text editor.
To do it with ldc the flag is `--output-s`. I have opened a PR to
make the ldc dmd-compatibility wrapper (`ldmd2`) mimic -vasm
Intel (and to a lesser extent Clang) actually annotate the
generated text with annotations intended to be read by the humans.
e.g.
Intel C++ (which is in the process of being replaced with Clang
relabeled as Intel C++) prints it's (hopeless unless you are
using PGO, but still) estimates of the branch probabilities.
```
test al, al#5.8
je..B1.4# Prob 22% #5.8
# LOE rbx rbp r12 r13 r14 r15
# Execution count [7.80e-01]
```
You can also ask the compiler to generate an optimization report
inline with the assembly code. This *is* useful when tuning since
you can tell what the compiler is or isn't getting right (e.g.
find which roads to force the loop unrolling down). The Intel
Compiler also has a reputation for having an arsenal of dirty
tricks to make your code "faster" which it will deploy on the
hope that you (say) don't notice that your floating point numbers
are now less precise.
`-qopt-report-phase=vec` yields:
```
# optimization report
# LOOP WITH UNSIGNED INDUCTION VARIABLE
# LOOP WAS VECTORIZED
# REMAINDER LOOP FOR VECTORIZATION
# MASKED VECTORIZATION
# VECTORIZATION HAS UNALIGNED MEMORY REFERENCES
# VECTORIZATION SPEEDUP COEFFECIENT 3.554688
# VECTOR TRIP COUNT IS ESTIMATED CONSTANT
# VECTOR LENGTH 16
# NORMALIZED VECTORIZATION OVERHEAD 0.687500
# MAIN VECTOR TYPE: 32-bits integer
vpcmpuq k1, zmm16, zmm18, 6 #5.5
vpcmpuq k0, zmm16, zmm17, 6 #5.5
vpaddqzmm18, zmm18, zmm19 #5.5
vpaddqzmm17, zmm17, zmm19 #5.5
kunpckbw k2, k0, k1#5.5
vmovdqu32 zmm20{k2}{z}, ZMMWORD PTR [rcx+r8*4] #7.9
vpxordzmm21{k2}{z}, zmm20, ZMMWORD PTR [rax+r8*4] #7.9
vmovdqu32 ZMMWORD PTR [rcx+r8*4]{k2}, zmm21 #7.9
add r8, 16#5.5
cmp r8, rdx #5.5
jb..B1.15 # Prob 82% #5.5
```
People don't seem to care about SPEC numbers too much anymore,
but the Intel Compilers still have many features for gaming
standard test scores.
http://www.spec.org/cpu2006/results/res2007q3/cpu2006-20070821-01880.html If
you looked at this, you'd think that Intel just managed a huge increase on
`libquantum` which we can all use on our own code, but it turns out they worked
out they can just tell the compiler to automagically parallelize the code, but
still only have 1 nominal process.
https://stackoverflow.com/questions/61016358/why-can-gcc-only-do-loop-interchange-optimization-when-the-int-size-is-a-compile
for more overfitting.
Compilers that take a detour through an assembler to generate
code are inherently slower.
Certainly, although in my experience not by much. Time spent in
the assembler in dominated by time spent in the linker, and just
about everywhere else in the compiler (especially when you turn
optimizations on). Hello World is about 4ms in the assembler on
my machine.
GCC and Clang have very different architectures in this regard
but end up being pretty similar in terms of compile times. The
linker an exception to that rule of thumb, however, in that the
LLVM linker is much faster than any current GNU offering.
It doesn't have a distinct IR like LLVM does but the final
stage of the RTL is basically a 1:1 representation of the
instruction set:
That looks like intermediate code, not assembler.
It is the (final) intermediate code, but it's barely intermediate
at this stage i.e. these are effectively just the target
instructions printed with LISP syntax.
It's, helpfully, quite obfuscated unfortunately: Some of that is
technical baggage, some of it is due to the way that GCC was
explicitly directed to be difficult to consume).
I'm __not__ suggesting any normal programmer should use, just
showing what GCC does since I mentioned LLVM.
Anyway, I've been playing with -vasm and I think it seems pretty
good so far. There are some formatting issues which shouldn't be
hard to fix at all (this is why we asked for some basic tests of
the shape of the
Re: DMD now incorporates a disassembler
On 09/01/2022 4:01 PM, Walter Bright wrote: I buried my PDP-11 long ago. Sob. There is a kit for the control panel[0]. Backed by a raspberry pi. I'm pretty keen to eventually buy one and build it. These kits are cool! [0] https://www.tindie.com/products/obso/pdp-11-replica-kit-the-pidp-11/
Re: DMD now incorporates a disassembler
On 1/7/2022 7:25 PM, Brian Callahan wrote: Thanks Walter. This is quite useful. Welcs. I'm already productively using it myself.
Re: DMD now incorporates a disassembler
On 1/7/2022 4:43 PM, Elronnd wrote: On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: : 89 F8 mov EAX,EDI Feature request: octal. I buried my PDP-11 long ago. Sob.
Re: DMD now incorporates a disassembler
On 1/7/2022 10:39 PM, ag0aep6g wrote: With feature creep in full swing now, when can I expect to read my email with DMD? The real question is why doesn't your email reader have an option to disassemble the email?
Re: DMD now incorporates a disassembler
On 1/8/2022 12:50 PM, max haughton wrote: On Saturday, 8 January 2022 at 18:47:11 UTC, Vladimir Marchevsky wrote: On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: Compile with -vasm to see it! Enjoy! Any practical reason to put disassembler into compiler instead of making it a separate tool? Any ETA for renaming it into DMD Burning ROM? :) Most other compilers have been able to do this for years. I've never seen one. What's the switch for gcc to do the same thing? The only difference is that the way the dmd backend is designed basically means that it never knows the instructions in a given basic block until they are actually emitted, so it has to disassemble it's own output rather than printing it's internal representation with (say) Intel assembly syntax. See https://llvm.org/doxygen/classllvm_1_1MachineInstr.html from LLVM GCC actually *only* uses an assembler to build object files. Compilers that take a detour through an assembler to generate code are inherently slower. It doesn't have a distinct IR like LLVM does but the final stage of the RTL is basically a 1:1 representation of the instruction set: That looks like intermediate code, not assembler.
Re: DMD now incorporates a disassembler
On 1/8/2022 10:47 AM, Vladimir Marchevsky wrote: On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: Compile with -vasm to see it! Enjoy! Any practical reason to put disassembler into compiler instead of making it a separate tool? Any ETA for renaming it into DMD Burning ROM? :) https://www.digitalmars.com/ctg/obj2asm.html
Re: DMD now incorporates a disassembler
On Saturday, 8 January 2022 at 18:47:11 UTC, Vladimir Marchevsky
wrote:
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote:
Compile with -vasm to see it! Enjoy!
Any practical reason to put disassembler into compiler instead
of making it a separate tool? Any ETA for renaming it into DMD
Burning ROM? :)
Most other compilers have been able to do this for years. The
only difference is that the way the dmd backend is designed
basically means that it never knows the instructions in a given
basic block until they are actually emitted, so it has to
disassemble it's own output rather than printing it's internal
representation with (say) Intel assembly syntax.
See https://llvm.org/doxygen/classllvm_1_1MachineInstr.html from
LLVM
GCC actually *only* uses an assembler to build object files. It
doesn't have a distinct IR like LLVM does but the final stage of
the RTL is basically a 1:1 representation of the instruction set:
```d
void phoneHome(size_t);
auto getLen(int[] arr)
{
phoneHome(arr.length);
return arr.length;
}
```
ends up as
```
;; Function getLen (_D7example6getLenFAiZm, funcdef_no=0,
decl_uid=1395, cgraph_uid=2, symbol_order=1)
(note 1 0 37 NOTE_INSN_DELETED)
(note 37 1 8 (var_location arr (parallel [
(expr_list:REG_DEP_TRUE (reg:DI 5 di [ arr ])
(const_int 0 [0]))
(expr_list:REG_DEP_TRUE (reg:DI 4 si [ arr+8 ])
(const_int 8 [0x8]))
])) NOTE_INSN_VAR_LOCATION)
(note 8 37 7 2 [bb 2] NOTE_INSN_BASIC_BLOCK)
(note 7 8 26 2 NOTE_INSN_FUNCTION_BEG)
(insn/f:TI 26 7 27 2 (set (mem:DI (pre_dec:DI (reg/f:DI 7 sp)) [0
S8 A8])
(reg:DI 3 bx)) "/app/example.d":2:6 54 {*pushdi2_rex64}
(expr_list:REG_DEAD (reg:DI 3 bx)
(nil)))
(note 27 26 2 2 NOTE_INSN_PROLOGUE_END)
(insn 2 27 38 2 (set (reg:DI 3 bx [orig:85 arr ] [85])
(reg:DI 5 di [92])) "/app/example.d":2:6 80
{*movdi_internal}
(nil))
(note 38 2 13 2 (var_location arr (reg:TI 3 bx [orig:85 arr ]
[85])) NOTE_INSN_VAR_LOCATION)
(call_insn:TI 13 38 39 2 (call (mem:QI (symbol_ref:DI
("_D7example9phoneHomeFmZv") [flags 0x41] 0x7f491d317600 phoneHome>) [0 phoneHome S1 A8])
(const_int 0 [0])) "/app/example.d":4:14 886 {*call}
(expr_list:REG_CALL_ARG_LOCATION (nil)
(expr_list:REG_DEAD (reg:DI 5 di)
(expr_list:REG_CALL_DECL (symbol_ref:DI
("_D7example9phoneHomeFmZv") [flags 0x41] 0x7f491d317600 phoneHome>)
(nil
(expr_list:DI (use (reg:DI 5 di))
(nil)))
(note/c 39 13 17 2 (var_location arr (nil))
NOTE_INSN_VAR_LOCATION)
(insn 17 39 36 2 (set (reg/i:DI 0 ax)
(reg:DI 3 bx [orig:85 arr ] [85])) "/app/example.d":6:1
80 {*movdi_internal}
(expr_list:REG_DEAD (reg:DI 3 bx [orig:85 arr ] [85])
(nil)))
(note 36 17 29 2 NOTE_INSN_EPILOGUE_BEG)
(insn/f 29 36 40 2 (set (reg:DI 3 bx)
(mem:DI (post_inc:DI (reg/f:DI 7 sp)) [0 S8 A8]))
"/app/example.d":6:1 62 {*popdi1}
(expr_list:REG_CFA_ADJUST_CFA (set (reg/f:DI 7 sp)
(plus:DI (reg/f:DI 7 sp)
(const_int 8 [0x8])))
(nil)))
(note 40 29 18 2 (var_location arr (nil) [uninit])
NOTE_INSN_VAR_LOCATION)
(insn 18 40 30 2 (use (reg/i:DI 0 ax)) "/app/example.d":6:1 -1
(nil))
(jump_insn:TI 30 18 33 2 (simple_return) "/app/example.d":6:1 910
{simple_return_internal}
(nil)
-> simple_return)
(barrier 33 30 25)
(note 25 33 0 NOTE_INSN_DELETED)
```
just prior to spitting it out for the assembler to process.
Re: DMD now incorporates a disassembler
On Sat, Jan 08, 2022 at 08:29:20PM +, max haughton via Digitalmars-d-announce wrote: > On Saturday, 8 January 2022 at 18:08:27 UTC, Steven Schveighoffer wrote: > > On 1/8/22 12:23 PM, jmh530 wrote: > > > On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: > > > > Compile with -vasm to see it! Enjoy! [...] > > > Would make a nice project for someone to integrate this into > > > run.dlang.org > > > > Isn't there already an ASM button? [...] > Yup. Better yet, the ASM button on run.dlang.org shows disassembly for all 3 compilers, not just dmd. T -- The early bird gets the worm. Moral: ewww...
Re: DMD now incorporates a disassembler
On Saturday, 8 January 2022 at 18:08:27 UTC, Steven Schveighoffer wrote: On 1/8/22 12:23 PM, jmh530 wrote: On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: Compile with -vasm to see it! Enjoy! [snip] Would make a nice project for someone to integrate this into run.dlang.org Isn't there already an ASM button? -Steve Yup.
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: Compile with -vasm to see it! Enjoy! Any practical reason to put disassembler into compiler instead of making it a separate tool? Any ETA for renaming it into DMD Burning ROM? :)
Re: DMD now incorporates a disassembler
On 1/8/22 12:23 PM, jmh530 wrote: On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: Compile with -vasm to see it! Enjoy! [snip] Would make a nice project for someone to integrate this into run.dlang.org Isn't there already an ASM button? -Steve
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: Compile with -vasm to see it! Enjoy! [snip] Would make a nice project for someone to integrate this into run.dlang.org
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote:
Compile with -vasm to see it! Enjoy!
For the file test.d:
int demo(int x)
{
return x * x;
}
Compiling with:
dmd test.d -c -vasm
prints:
_D4test4demoFiZi:
: 89 F8 mov EAX,EDI
0002: 0F AF C0imulEAX,EAX
0005: C3 ret
https://github.com/dlang/dmd/pull/13447
Great news, this worth a hn / reddit post!
Kind regards
Andre
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote:
Compile with -vasm to see it! Enjoy!
For the file test.d:
int demo(int x)
{
return x * x;
}
Compiling with:
dmd test.d -c -vasm
prints:
_D4test4demoFiZi:
: 89 F8 mov EAX,EDI
0002: 0F AF C0imulEAX,EAX
0005: C3 ret
https://github.com/dlang/dmd/pull/13447
Nice!
Re: DMD now incorporates a disassembler
On Sat, Jan 08, 2022 at 07:39:54AM +0100, ag0aep6g via Digitalmars-d-announce
wrote:
> On 07.01.22 22:41, Walter Bright wrote:
> > Compile with -vasm to see it! Enjoy!
>
> With feature creep in full swing now, when can I expect to read my email
> with DMD?
You already can:
echo 'import std;void main(){execute("/usr/bin/mail");}' | dmd -run -
:-P
T
--
"Real programmers can write assembly code in any language. :-)" -- Larry Wall
Re: DMD now incorporates a disassembler
On 07.01.22 22:41, Walter Bright wrote: Compile with -vasm to see it! Enjoy! With feature creep in full swing now, when can I expect to read my email with DMD?
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote:
Compile with -vasm to see it! Enjoy!
For the file test.d:
int demo(int x)
{
return x * x;
}
Compiling with:
dmd test.d -c -vasm
prints:
_D4test4demoFiZi:
: 89 F8 mov EAX,EDI
0002: 0F AF C0imulEAX,EAX
0005: C3 ret
https://github.com/dlang/dmd/pull/13447
Thanks Walter. This is quite useful. Will put it through its
paces. Already spotted some print formatting weirdness; will send
as bug reports.
~Brian
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: : 89 F8 mov EAX,EDI Feature request: octal.
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote:
Compile with -vasm to see it! Enjoy!
For the file test.d:
int demo(int x)
{
return x * x;
}
Compiling with:
dmd test.d -c -vasm
prints:
_D4test4demoFiZi:
: 89 F8 mov EAX,EDI
0002: 0F AF C0imulEAX,EAX
0005: C3 ret
https://github.com/dlang/dmd/pull/13447
Wow, very useful! This feature surely lowers the bar to check the
disassembly when optimising. Thanks!
I'm slightly disappointed it does not output the asm inlined to D
code but that's just my daydreaming with no practical reasons to
back it up.
Re: DMD now incorporates a disassembler
On Friday, 7 January 2022 at 21:41:55 UTC, Walter Bright wrote: Compile with -vasm to see it! Enjoy! Oh very nice! I need to finish my debugger/disassembler project.
DMD now incorporates a disassembler
Compile with -vasm to see it! Enjoy!
For the file test.d:
int demo(int x)
{
return x * x;
}
Compiling with:
dmd test.d -c -vasm
prints:
_D4test4demoFiZi:
: 89 F8 mov EAX,EDI
0002: 0F AF C0imulEAX,EAX
0005: C3 ret
https://github.com/dlang/dmd/pull/13447
